Design an intelligent courtroom simulation system using Large Language Model (LLM) agents. Given a textual case description, your system must simulate a full courtroom trial involving multiple autonomous agents, each representing key legal roles.

Minimum Required Agents:
● Defendant
● Defense Lawyer
● Plaintiff
● Prosecution Lawyer
● Judge

Phases of the Simulation:
1. Opening Statements
○ Defense and Prosecution introduce their sides of the case.
2. Witness Interrogation & Argumentation
○ Defense and Prosecution can each interrogate witnesses or construct iterative arguments.
3. Closing Statements
○ Each side gives a final summary of their position.
4. Judge’s Ruling (Final Phase)
○ The judge agent deliberates and delivers a verdict based on the arguments, evidence, and witness testimonies.

Bonus Objectives (Extra Credit):
● Support freely evolving trial structure —
Even though you are provided with a basic conversation structure, you may implement flexible trial progression based on how the case unfolds, ensuring it is grounded in reality.
● Implement dynamic agent creation —
You may extend this system with additional roles such as: 
○ Witnesses (called during the trial)
○ Jury members
○ Expert consultants
○ Bailiff, stenographer, etc.

These roles should be dynamic — agents can be spawned or dismissed based on the needs of the case.